Abstract
This thesis presents a possible framework to evaluate and mitigate the technical impacts of large-scale electric vehicle (EV) charging on medium-voltage (MV) and low-voltage (LV) distribution networks. To that end, realistic test feeders were constructed in DIgSILENT PowerFactory across radial, mesh, and star topologies, using standardized network parameters. Agent-based modeling was used to generate residential electricity demand profiles, reflecting diverse demographic and behavioral attributes across typical European households.These were integrated with probabilistic EV charging scenarios, derived from empirical mobility data, to simulate uncontrolled Level 2 home charging across varying adoption levels and spatial distributions. Power flow simulations reveal that under 50% EV penetration, radial LV networks experience voltage drops exceeding -8.2% and cable loading surpassing 120% of thermal limits during evening peaks. Transformer loading exceeds 1.2 p.u. in dense neighborhoods, indicating capacity violations under unmanaged charging. While distributed solar generation provides partial relief, it is often misaligned with peak evening charging demand, limiting its effectiveness without storage or control mechanisms.
To mitigate these issues, a spatial optimization framework was developed using neural network surrogates trained on full AC load flow outputs. Applied across test networks, this strategy reduces voltage violations by up to 70% and peak cable loading by 38% in worst-case scenarios. The study further evaluates two hardware-level interventions: mobile and fixed battery energy storage systems, and a vertically stacked low-cost stationary storage solution. Combined, these systems reduce transformer overloading by 45–60% and shift approximately 22% of peak EV charging load to unconstrained hours.
To support deployment at scale, a peer-to-peer (P2P) transactional business model is proposed, linking locational grid stress to dynamic charging prices. Simulations demonstrate that this model lowers localized overload probabilities by 26% and improves overall load flattening by 19%. The combined technical and economic framework enables high EV penetration without extensive grid upgrades, offering a scalable pathway to electrified transport aligned with distribution network resilience.
| Date of Award | 16 Jun 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Victor Becerra (Supervisor) & Anton Hettiarachchige Don (Supervisor) |